Computers can perform image processing using complex computer vision (CV) schemes, such as convolutional neural networks. The CV schemes can be computationally intensive and usually require powerful processors and large amounts of memory. It is often impractical to execute CV schemes on mobile devices because the mobile devices have limited computational resources, which causes long processing times. More so, applying an intensive CV scheme in real time (e.g., 30 frames per second) using a mobile device is currently very difficult, if not impossible.